Limited hybridisation and introgression despite stocking among endemic Interior Highlands black basses (Centrarchidae: Micropterus)

Smallmouth Bass (Micropterus dolomieu; SMB) are globally popular among anglers and have been widely introduced (i.e. stocked) for population management and sportfishing. Importantly, stocking was prevalent before cryptic diversity within the SMB complex was known, which now includes three newly elevated species: Neosho Bass (M. velox; NB), Little River Bass (M. sp. cf. dolomieu Little River; LRB) and Ouachita Bass (M. sp. cf. dolomieu Ouachita River; OB). We sought to quantify population structure and hybridisation and introgression in these three recently described species.


| INTRODUC TI ON
The first crucial steps in biological conservation are to identify lineages with evolutionarily significant variation (i.e.biodiversity), and then to understand the processes both generating and threatening diversity (Garrick et al., 2019;Jansson & Dynesius, 2002).Both steps are challenging if distinct taxa are cryptic, either due to recent diversification, plasticity or morphological convergence (Culver et al., 1995;Culver & Pipan, 2009;Schluter, 1996).Modern advancements in genetics have propelled the delimitation (and taxonomic elevation) of distinct lineages within species originally assumed to occupy large geographic ranges (e.g.Kim et al., 2021Kim et al., , 2022)).In many cases, these species may be range-restricted by physical or ecological mechanisms and exhibit locally high fitness under extreme environmental conditions (Bickford et al., 2007;Funk et al., 2012;Niemiller et al., 2012).Disruption of local adaptations could have implications for long-term population viability and overall community diversity.
The black basses (Centrarchidae: Micropterus) have been prized as sportfishes in North America for two centuries (Long et al., 2015; USFWS, 2016), supporting a billion-dollar recreational fishing industry (Long et al., 2015).Federal, state, and private entities have conducted relatively unrestrained and pervasive stocking (i.e.introduction of wild-caught or hatchery-reared fishes en masse into streams or lakes) since the late 1800s (Long et al., 2015) to maximise angling opportunities or as a means of restoring small populations at risk of local extinction.Stocking was originally prescribed at a time when the genus was thought to comprise only two species: Largemouth Bass (M.nigricans) and Smallmouth Bass (M.dolomieu; Henshall, 1881).However, a recent surge of genetic inquiry has revealed cryptic diversity that was not fully characterised by earlier investigations using comparative morphology.Molecular analyses have resolved fine scale phylogeographical boundaries within some black basses, such as in the Redeye Bass (M.coosae) complex (Baker et al., 2013), and have led to the discovery of cryptic species (e.g.Choctaw Bass, M. sp.cf.punctulatus; Long et al., 2015;Taylor et al., 2019;Tringali et al., 2015).
Recent black bass surveys have focused on the freshwater endemism hotspot in the Central Interior Highlands (CIH; Borden & Krebs, 2009;Gunn et al., 2020Gunn et al., , 2022;;Soltis et al., 2006;Stark & Echelle, 1998;Taylor, Long, et al., 2018) spanning the Ouachita Mountains, Boston Mountains, and the lower Ozark Highlands of North America.Researchers have paid special attention to the Smallmouth Bass given its commercial popularity, complex phylogeographical history, and known occurrence of introductions within and outside the species' native distribution.Hubbs and Bailey (1940) first described two subspecies, the Neosho Smallmouth Bass (M.d. velox), inhabiting tributaries of the Arkansas River in the Ozark Highlands and Boston Mountains, and the Northern Smallmouth Bass (M.d. dolomieu), encompassing the remainder of the range from the central United States (USA) to southern Canada east of the Great Plains (Brewer & Orth, 2015).Hubbs and Bailey (1940) also proposed two 'intergrade zones', one within the White, Black and St. Francis River basins of southern Missouri and northern Arkansas, and one between the Red and Ouachita rivers in the Ouachita Mountains of south-eastern Oklahoma and south-western Arkansas.In a survey of 33 allozyme loci across samples collected in the CIH and some sites in the north-eastern USA, Stark and Echelle (1998) reaffirmed distinctions between the subspecies but suggested the 'intergrades' in the Ouachita Mountains represented a third genetic lineage, the Ouachita Smallmouth Bass (M.sp.cf.dolomieu velox), itself containing two distinct allopatric populations in the Ouachita River and Little River basins.Since then, black bass lineages in the CIH have been distinguished morphologically (Gunn et al., 2020), genetically (Bagley et al., 2011;Boxrucker et al., 2004;Gunn et al., 2020;Long et al., 2021;Taylor, Long, et al., 2018), and with high-throughput genomic sequencing (Gunn et al., 2022;Kim et al., 2022) Genomic variation in the Smallmouth Bass species complex is concentrated in the CIH (Kim et al., 2022).Despite this, stocking in parts of the region (Boxrucker et al., 2004;Stark & Echelle, 1998) has occurred with likely fitness consequences.Hybridisation and subsequent introgression of non-native ancestry may have significant consequences for the survival and reproduction of a recipient population, either through increased hybrid fitness (i.e.heterosis) leading to genetic swamping (Avise, 2010), or through the disintegration of genomic regions underlying environmental adaptation (i.e.outbreeding depression; Altukhov & Salmenkova, 1987).
Hybridisation between Smallmouth Bass and Neosho Bass has been recently reported (Gunn et al., 2020(Gunn et al., , 2022;;Taylor, Long, et al., 2018), and extensive introgression has been attributed directly to stocking in upstream portions of the Illinois River following introduction of hatchery-reared Smallmouth Bass ('Tennessee Lake-strain', basin-restricted species in the Ouachita Mountains ecoregion that may be managed for long-term conservation.
Tennessee Lake-strain Smallmouth Bass were concurrently stocked in Broken Bow Lake, Oklahoma, within the native range of the Little River Bass (Boxrucker et al., 2004), and later into Lake Ouachita, Arkansas in 2004 (Port et al., 2021) within the native range of the Ouachita Bass, after stocking of native Ouachita Bass failed to produce a viable fishery in the reservoir.However, the relative newness of species delimitation and suspected but poorly documented historical stockings in the region suggest that some unreported introgression may exist in these species.
Ouachita Bass and Little River Bass have been isolated in separate basins in the Ouachita Mountains for millennia and are adapted to local ecosystem dynamics.Given their genomic divergence (Kim et al., 2022) and allopatric ranges, these species likely exhibit unique ecologies, for example, habitat requirements, diet, ecophysiology and behaviour.Experts have reviewed the life history traits of Smallmouth Bass (Brewer & Orth, 2015) and Neosho Bass (Brewer & Long, 2015); however, apart from one recent study finding elevated critical thermal maximum in Ouachita Bass compared to Neosho Bass (Brewer et al., 2022), differences across the entire Smallmouth Bass species complex have not been fully described.Known characteristics of the Ouachita Mountains ecoregion may contribute to ecological differentiation.The Ouachita Mountains are geologically and ecologically diverse and contain at least 10 endemic fishes (Mayden, 1985) that are confined to small runoff streams (Hafs et al., 2010) and thus subjected to extreme variability in abiotic (e.g.flow regimes, water temperatures and suitable habitat; Hines, 1975;Homan et al., 2005) and biotic (competition and predation;Gagen et al., 1998;Harvey & Stewart, 1991) conditions.

The presence of potentially strong local adaptation in the
Ouachita Mountains could mean that populations are vulnerable to the deleterious effects of outbreeding.However, data on hybridisation, introgression, and genetic structure are limited across the range.Admixture may be widespread in these species, either through anthropogenic (i.e.stocking) or presently unknown natural (e.g.periodic flooding, stream capture or ground water flow) mechanisms.Furthermore, global change, including rise in average water temperature and industrial development, may affect ecological systems in ways that create new opportunities for species, allopatric or sympatric, to hybridise.Little River Bass and Ouachita Bass are naturally sympatric with at least one closely related congener (Spotted Bass, M. punctulatus); while prezygotic reproductive isolation would be expected to minimise hybridisation between species evolved in natural sympatry, other species in the Smallmouth Bass species complex interbreed readily with Spotted Bass (Avise et al., 1997;Koppelman, 1994).Assessing levels of introgressive hybridisation is critical to understanding whether presumed local adaptations can overcome or protect against propagule pressure associated with mixing non-native forms.
We used a low-density diagnostic panel of 192 SNPs (Long et al., 2021) to conduct a range-wide genetic survey of the Smallmouth Bass species complex and to test for hybridisation and introgression due to stocking in the Little River Bass and Ouachita Bass in the CIH.
We included Smallmouth Bass, Neosho Bass (Gunn et al., 2020;Long et al., 2021;Taylor, Long, et al., 2018) and Spotted Bass (a closely related outgroup), to assess levels of natural and human-mediated hybridisation and introgression based on shared allelic composition.
We expected strong genetic structure differentiating Little River Bass and Ouachita Bass, as suggested by allozyme-based analyses conducted over 25 years ago by Stark and Echelle (1998).Finally, we hypothesised that some Little River Bass and Ouachita Bass stream populations would exhibit substantial introgression with non-native Smallmouth Bass, at least in areas with historic stocking.

| Sample collection
We collected fin clips from all major river drainages spanning the currently recognised native ranges of two species within the Smallmouth Bass species complex (SMB-C) in the Ouachita Mountains (Kim et al., 2022) 1a,b).
We included stream sites adjacent to all large impoundments, including Broken Bow Lake in Oklahoma, USA, and Lake Greeson, Lake DeGray and Lake Ouachita in Arkansas, USA; additionally, we included samples from one tributary of the Arkansas River basin (Black Fork of the Poteau River; Table 1; Figure 1b) that had been previously stocked with broodfish originating from the Little River basin (Stark & Echelle, 1998).
To assess hybridisation and introgression of OB and LRB with allopatric Neosho Bass (M.velox; NB) and introduced Smallmouth Bass (M.dolomieu; SMB), we used reference samples of SMB, which had been previously screened for non-native ancestry from Spotted Bass (M.punctulatus; SPB) and NB, which had been previously screened for non-native ancestry from SMB andSPB (2015-2018;Gunn et al., 2020;Long et al., 2021;Taylor, Long, et al., 2018).Briefly, NB samples were collected from five native streams in the Arkansas River basin whereas SMB samples were collected from three native lentic populations (Lake Superior, Lake Erie, and Oneida Lake, New York, USA), one native lotic population (Smith Fork Creek, Tennessee, USA), and one non-native impoundment in Oklahoma (Skiatook Lake) that was stocked in 1991-1992 with hatchery-reared fish from the Cumberland River drainage (Tennessee Lake-Strain; Figure 1b).
To assess natural hybridisation with a sympatric congener, we also obtained a genetically pure reference set of SPB from the Illinois River and Glover River in the Arkansas River basin (Table 1).For all samples, fish were captured by hook-and-line angling or boat electrofishing (Long et al., 2021), visually identified to species, and ~1 cm 2 of tissue from the upper caudal fin was removed with sterile clippers and stored in 95% ethanol at −20°C until processing.
TA B L E 1 Species, stream or lake 'population', site, latitude and longitude, and sample sizes (n) for 484 black bass collected from across the Interior Highlands, including Spotted Bass (Micropterus punctulatus; SPB), Smallmouth Bass (M.dolomieu; SMB), Neosho Bass (M.velox; NB), Little River Bass (M.sp.cf.dolomieu Little River; LRB) and Ouachita Bass (M.sp.cf.dolomieu Ouachita River; ORB).for each round was the number at which cumulative variance curves began to plateau, and one LD was retained for visualisation.Finally, we screened for duplicated individuals (e.g. if collected on multiple occasions) by calculating pairwise per cent identity between samples, and we omitted samples exceeding 95% identity.

| Hybridisation and population structure in the Smallmouth Bass species complex (SMB-C)
We estimated species-specific ancestry proportions using Bayesian clustering analysis in Structure v.2.3.4 (Pritchard et al., 2000), and we identified individuals of hybrid origin using Bayesian assignment analysis in NewHybridS v.1.1b(Anderson & Thompson, 2002).First, we used SPB-SNPS to infer ancestry and hybridisation between SPB and all species in the SMB-C, including the NB reference set.
To eliminate ancestry bias in downstream analyses, pure SPB and hybrids were then omitted from the dataset.Second, we used SMB-SNPS to infer ancestry and hybridisation between SMB and all other species in the CIH, including remaining NB reference individuals.For each analysis in Structure, we used the admixture model with no prior for location and ran 500,000 burn-in steps and 1,000,000 Markov chain Monte Carlo (MCMC) iterations for 10 replicates at each a priori number of clusters (K).Values of K were set based on the number of stream sites represented in the data.We determined optimal K using the deltaK metric (Evanno et al., 2005) implemented in R with the package pophelper v.2.3.1 (Francis, 2017).We additionally inferred optimal K using Puechmaille metrics (Puechmaille, 2016) implemented in the online software Structure Selector (Li & Liu, 2018); these metrics are especially useful when individuals can be a priori grouped into geographically meaningful 'subpopulations' (in our case, rivers), and they are robust to uneven sample sizes (Puechmaille, 2016).
For optimal K, we processed raw structure replicates in clumpp 100,000 burn-in steps and 500,000 sweeps.Resulting posterior probabilities of assignment to each hybrid category for all individuals were visualised in R. Individuals were identified as likely hybrids if they were assigned with greater than 50% posterior probability to any of four hybrid categories (F 1 , F 2 , Backcross to Parent 1, Backcross to Parent 2); this threshold was chosen based on a previous power analysis showing relatively high power to detect hybrids regardless of category or number of SNP loci (Long et al., 2021).

| Introgression in the Smallmouth Bass species complex (SMB-C)
We further assessed the extent of allelic introgression in hybrids

| Hybridisation, introgression and population structure in the Smallmouth Bass species complex (SMB-C)
Scanning the full dataset for ancestry proportions and hybridisation between SPB and species in the SMB-C (SPB-SNPS), we found strong support for K = 2 genetic clusters using both del-taK (K = 12,975.06;Table S1) and Puechmaille (MedMedK = 2, MedMeanK = 2, MaxMedK = 2, MaxMeanK = 2; Table S2) metrics, with SPB individuals assigning mostly to one cluster (q ave = 0.99) and SMB-C individuals assigning mostly to a second cluster (q ave = 0.98; Figure 3a; Figure S2a).We found no evidence of SPB introgression in SMB sites (Table 2).Conversely, we detected 26 multigeneration hybrids (frequency across species of ~0.06) distributed across NB (n = 2), LRB (n = 22), and OB (n = 2) streams (Table 2), all of which were either backcrossed to SMB-C or F 1 hybrids.In NB, we found a single SMB-C backcross in Honey Creek and a single F 1 hybrid in Lee Creek.In the LRB native range, most hybrids were found in Glover River (SMB-C backcrosses, n = 5) and Western Saline River (SMB-C backcrosses, n = 9).Additional SMB-C backcrosses were found in the Little River (n = 1), Upper Mountain Fork (n = 2), and the mainstem of Mountain Fork (n = 2).We found one F 1 hybrid in Blackfork Creek.
In OB, we found two SMB-C backcrosses in Eastern Saline River (Table 2; Figure S2a).
After removing SPB hybrids from the full dataset, we assessed hybridisation between SMB and all other CIH species (SMB-SNPS).We found strong support for K = 3 using the deltaK metric (K = 1042.10;Table S3) and K = 4 using Puechmaille metrics (MedMedK = 4, MedMeanK = 4, MaxMedK = 4, MaxMeanK = 4; Table S4).Individual ancestry proportions were highly variable among NB, LRB and OB populations in the CIH for both K = 3 and K = 4, with no consistent clustering patterns (Figure S3a,b).We therefore visualised ancestry proportions at K = 2.All SMB individuals confidently assigned to a single cluster (q ave = 0.99), and all CIH individuals assigned mostly to another cluster (q ave = 0.98; Figure 3b; Figure S2b).We found 22 SMB hybrids (~0.05) distributed across NB (n = 5), LRB (n = 15), and OB (n = 2), all of which were either backcrossed to CIH or F 2 hybrids (Table 3).Again, most hybrids in the LRB were in Upper Mountain Fork (F 2 , n = 3; CIH backcrosses, n = 5) and in the mainstem of Mountain Fork (F 2 , n = 1; CIH backcrosses, n = 6).In the NB range, we found one CIH backcross in Honey Creek, three CIH backcrosses in Spavinaw Creek and one CIH backcross in Baron Fork.Both hybrids in the OB native range were detected in the Ouachita River (Table 3; Figure S2b).
After removing SMB from the dataset and assessing hybridisation among CIH species (CIH-SNPS), we found support for K = 3 genetic clusters using deltaK (K = 24,187.47;Table S5) and K = 8 genetic clusters using Puechmaille metrics (MedMedK = 8, MedMeanK = 8, MaxMedK = 8, MaxMeanK = 8; Table S6).At K = 3, individuals clustered by species, with all NB individuals assigning mostly to one cluster (q ave = 0.98), all LRB individuals assigning mostly to a second cluster (q ave = 0.99), and all OB individuals assigning mostly to a third cluster (q ave = 0.99; Figure 3c; Figure S2c).We found a single LRB backcross (hybridised with OB) in the Western Saline River in the LRB range (data not shown; Figure S2c).At K = 8, patterns of ancestry were well resolved for some clusters, with OB individuals in Little Missouri River, Caddo River, and Ouachita River assigning mostly to one cluster (q ave = 0.97) and OB individuals in Eastern Saline River assigning to another cluster (q ave = 0.95), with some allele sharing between these two clusters in Eastern Saline River.NB individuals also largely assigned to a single cluster (q ave = 0.91) except for Lee Creek, which was not well-resolved.In the LRB range, Blackfork Creek individuals assigned largely to a single cluster (q ave = 0.86), and Western Saline River individuals assigned largely to another cluster (q ave = 0.74), but neither site was well-resolved.All other stream sites in the LRB range exhibited highly variable proportions of ancestry from two distinct clusters along with some allele sharing with other LRB clusters (Figure 3d; Figure S4).).'Species Totals' give total fish for each parental or H category within species.'Range-wide Totals' give overall total fish for each parental or H category across all species.Blank cells indicate a value of '0', but zeros are explicitly listed in the 'Species Totals' row for each species.
For hybrids between SMB and species native to the CIH, hybrid index varied between 0.45-1.00,with values closer to 1.00 indicating greater allelic contribution from CIH. Interspecific heterozygosity varied from 0.00 to 0.39 (Figure 4b).Inferred F 2 hybrids had different hybrid index values (0.62-1.00) and variable interspecific heterozygosity (0.03-0.36; Figure 4b).).'Species Totals' give total fish for each parental or H category within species.'Range-wide Totals' give overall total fish for each parental or H category across all species.Blank cells indicate a value of '0', but zeros are explicitly listed in the 'Species Totals' row for each species.has been a practice for nearly 200 years (Long et al., 2015) and with limited documentation, past stocking events cannot be completely catalogued, but our data suggest any stocking events that may have occurred were rarely successful within streams.In situations where stocking was maximised for success, such as in reservoirs with lenticadapted populations (e.g.Broken Bow Lake and Ouachita Lake), we detected SMB introgression in the streams directly connected to those reservoirs.As a result, the distinct genomic composition of each LRB and OB species remains relatively intact.
Introgression of SPB ancestry in the CIH was surprising.SPB introgression was extensive in the LRB range, most notably in the Little River (20% hybrids), Glover River (~10%), Mountain Fork (~13%) and Western Saline River (~24%; Figure 3a).Hybridisation is common in nature (Schumer et al., 2016) and occurs among black basses (Koppelman, 2015), but it has mostly been observed in black basses when non-native populations have been stocked.When stocked outside their indigenous range, non-native SPB introgression into a SMB genomic background can be substantial (up to 48%) depending on environmental conditions (Koppelman, 1994(Koppelman, , 2015)).
However, natural hybridisation is thought to be relatively rare in sympatric species (e.g.Gunn et al., 2020).We are unaware of studies documenting hybridisation among sympatric species at the levels we detected.
Heterogeneous admixture with SPB and the presence of SMB-C backcrosses in the Ouachita Mountains suggest that hybridisation is unabated by postzygotic barriers, as is typical for the genus (Koppelman, 2015).We found only two F 1 hybrids and no SPB backcrosses, and we found highly variable interspecific heterozygosity, likely indicating a range of earlier-(high interspecific heterozygosity) to later-generation (low interspecific heterozygosity) hybrids.
These patterns suggest that native SMB-C genomic ancestry may be favoured in the CIH.Although limited, hybridisation and introgression in these systems appear to be facilitated by unknown forces.
The Glover and Western Saline rivers are categorised by highly flashy flow regimes, with periods of limited to no water flow lasting 2-50 days per year (Leasure et al., 2016).Such conditions could facilitate interbreeding if the two species intermittently co-occur in abnormally high densities (Bolnick, 2009).Recent surveys in the Glover River quantified co-occurrence levels for LRB and SPB, with SPB density outnumbering LRB by a factor of 2.75 in the middle reaches of the stream, but greater density of LRB in lower stream reaches by a factor of 9.5 (Starks et al., 2017).The Glover River's flashy flow regime has been implicated in LRB recruitment failure (Brewer & Long, 2015), and our results provide some evidence it may also influence admixture between sympatric black basses.For instance, during spawning, flashy systems would favour spawning in habitats less susceptible to water level fluctuations, such as pools, potentially clustering nests closer together, which has been observed for NB (Miller & Brewer, 2021).In sunfishes (Lepomis), sister group to the black basses, nest clustering has been found to foster hybridisation (Jennings & Philipp, 2002).Elevated rates of mixing could be associated with climate change-induced increases in flow variability (Taylor et al., 2019), but background rates of hybridisation in these systems do not currently exist for testing.et al., 2006).For black basses, mate recognition includes visual cues through colour changes and increased contrast with dark (Enriquez et al., 2016).Turbidity can influence the expression of certain colours in fish (Atkinson, 2016), potentially affecting mate recognition.Although the influence of turbidity on hybridisation has not been assessed directly, land-use has been shown to correlate with rates of hybridisation in Bartram's Bass (M. sp. cf. coosae;Peoples et al., 2021).For this species, instances of pure Bartram's Bass were more prevalent in streams with higher proportions of forested land Incidences of non-native SMB introgression in the CIH were limited to watersheds where stocking has occurred, especially in reservoirs (Lake Tenkiller, Broken Bow Lake and Ouachita Lake), in line with expectations in other situations (Gunn et al., 2020(Gunn et al., , 2022;;Taylor, Long, et al., 2018).We intentionally focused our sampling efforts on portions of mainstem tributaries upstream of reservoirs to maximise the probability of locating fish with native ancestry.We found evidence of introgression several kilometres upstream of Lake Tenkiller in the Illinois River system (Honey Creek, Spavinaw Creek, Caney Creek, and Baron Fork; NB native range); upstream of the Mountain Fork system (upper Mountain Fork [~24%] and the mainstem river [~54%]; LRB range); and upstream of Ouachita Lake in the Ouachita River (~5%, OB native range; Figure 3b).However, the frequency of introgression was not consistent among sites within systems.Caney Creek, a small tributary of the Illinois River, showed no signature of admixture, compared with hybridisation frequencies of 11.11%, 33.33% and 10.00% in Honey Creek, Spavinaw Creek, and Baron Fork Creek, respectively, despite being nearly equidistant from Lake Tenkiller.We hypothesise that stream size may be a natural barrier to introgression from non-native SMB, whose origins generally include larger rivers as natural habitat (i.e.Tennessee lakestrain SMB; Taylor, Long, et al., 2018).Sampling additional tributaries to reservoirs where Tennessee lake-strain SMB have been stocked (e.g.Blakely Creek, Irons Fork, South Fork and North Fork inflowing to Ouachita Lake) would help test hypotheses of native CIH SMB-C species to be better adapted to small streams.SMB-C species in the CIH are considered fluvial-specialists (Brewer & Long, 2015), intolerant of reservoir conditions.In fact, it was lack of SMB in these reservoirs that prompted state agencies to stock non-native 'strains' to establish populations for angling (Boxrucker et al., 2004;Stark & Echelle, 1998).
The relative lack of SMB hybrids in the LRB and OB native ranges, given a history of stocking in the region and a propensity for interspecific hybridisation among black basses, could be explained by several factors.Ecoregional boundaries formed over millennia are likely responsible for speciation and population structuring in the CIH (Kim et al., 2022;Mayden, 1985) and limit introgression of non-native alleles.In fact, the geography of the Ouachita Mountains may help reduce spread of introduced SMB and their hybrids, because patches of suitable habitat in upland streams are connected only at points further downstream in the lowlands of the area, which is largely unsuitable for SMB-C species.Another possibility is that native LRB and OB may outcompete non-native SMB in high gradient fluvial habitats in the Ouachita Mountains (i.e.small streams) compared to low gradient habitats like large rivers and impoundments.In fact, there is mounting genomic evidence of local adaptation across the CIH (Gunn et al., 2022) and significant physiological differences, such as increased thermal tolerance in the lineages occupying the Ouachita Mountains (Brewer et al., 2022).Full descriptions of the morphological, physiological and ecological differences among species will be critical for identifying adaptations in the CIH and understanding their role in mediating interspecific gene flow.It is also important to note possible ascertainment bias derived from specimen sampling and SNP panel design.We intentionally collected fish from the putative native ranges of each species and based on external morphology.Although we did not exclude fin clips from fish with marginal variations in colour and patterning, we may have excluded some individuals with pronounced yet unknown hybrid phenotypes which may appear in earlier generation hybrid offspring.Additionally, the SNP panel used for genotyping (Long et al., 2021) is nuanced since it was constructed with a separate, limited set of fishes, and it therefore may not represent the full breadth of genomic variation among black bass species in the CIH.Both sampling strategy and panel design could have resulted in underestimation of hybrid frequencies.
Lack of widespread introgression with SMB in the CIH requires careful consideration and a precautionary approach to additional stocking in the region.Many studies point to habitat alteration and propagule pressure as dynamic mechanisms that can influence rates and outcomes of species invasions and introgression events (e.g.Bangs et al., 2016;Bennett et al., 2010;Lockwood et al., 2005;Pringle et al., 2000;von Holle & Simberloff, 2005).Even among black basses, anthropogenic habitat alteration can increase introgression rates (Koppelman, 1994).Sources of non-native propagules, such as impoundments stocked with non-native black basses, have also been implicated in increased invasion and introgression rates of black basses across riverscapes (Judson et al., 2021;Peoples et al., 2021;Taylor, Tringali, et al., 2018).Such mechanisms, along with hydrological changes associated with climate change in the region, could alter introgression rates from their observed rates in the present study.
Fine-scale sampling and monitoring in systems with known sources of non-natives (e.g.Broken Bow Lake in the LRB range) is warranted, along with establishing genetic management units that could help conserve remaining genetic biodiversity within each member of the species complex.

| Conclusions
Our findings represent an important baseline of biodiversity within the SMB-C, and the first such study focused on the LRB and OB.The presence of some recent hybrids (i.e.F 1 and F 2 ) and early-and later-generation backcrosses in populations of LRB and OB confirms that species in the SMB-C can readily hybridise and may be susceptible to the homogenising effects of gene flow.As basin-restricted species, they may be especially vulnerable to the effects of climate and environmental change, which is not unexpected given that nearly all endemic fishes in the Ouachita Mountains ecoregion are threatened or imperilled (Jelks et al., 2008).Yet despite known stocking of SMB in reservoirs throughout the Ouachita Mountains and more broadly across the CIH, introgression of non-native alleles is generally low.If adaptations exist to facilitate maintenance of native alleles in these species, then the effects of past stocking could be ameliorated.
Stocking records are sparse, especially in the 1800s when black bass taxonomy was poorly understood and stocking was widespread (Long et al., 2015).The degree to which possible historical stockings have influenced genomic variation is largely unknown.Furthermore, our SNP panel was designed to diagnose SMB-C species (Long et al., 2021) and detect allele-sharing but is not appropriate for estimating population genetic diversity metrics or identifying genes under selection.This level of information would be useful to understand the demographic history of populations, specifically whether past stockings failed to establish individuals or if those individuals contained maladapted alleles.

Fin
clips were sent to the Center for Aquaculture Technologies (CAT) for DNA extraction, quality assessment and sequencing.DNA extractions were performed on ~15 mg pieces of tissue using a standardised magnetic bead-based extraction protocol (CAT), and eluted samples were arranged randomly on 96-well plates.For quality control, 10-12 samples were selected from each plate, quantified on a Nanodrop spectrophotometer and screened for low yield and gDNA degradation on a 2% agarose gel (50-100 μL at ~30 ng/μL).DNA samples were amplified following established guidelines using Kompetitive Allele Specific Polymerase Chain Reaction (KASP) technology (LGC Genomics).All samples were genotyped on a lowdensity SNP panel designed to determine allelic ancestry and multigeneration hybrids among black bass species in the SMB-C and CIH (Figure 2a).The panel consisted of 192 SNPs derived from restriction enzyme associated DNA sequencing (RADseq) of 76 black bass individuals representing SPB, SMB, NB, OB and LRB.Three distinct sets of SNPs were selected, each maximising genomic differentiation among species on three hierarchical levels (Long et al., 2021): (1) SPB versus all four species in the SMB-C (SPB-SNPS; Figure 2b); (2) SMB versus all species native to the CIH (SMB-SNPS; Figure 2c); and (3) NB versus OB versus LRB across basins in the CIH (CIH-SNPS; Figure 2d).
analysis of principal components (DAPC; Jombart et al., 2010) in R v.4.2.2 (R Core Team, 2022) through the package adegenet v.1.3.1 (Jombart, 2008; Jombart & Ahmed, 2011).To detect bias in each hierarchical analysis (SPB vs. SMB-C with SPB-SNPS; SMB vs. CIH with SMB-SNPS; CIH with CIH-SNPS), we compared linear discriminant (LD) loadings of genetic groups from two rounds of DAPC, one for all data including missing genotypes, and one for all data excluding missing genotypes.Number of principal components (PCs) retained SMB and inferred hybrids were then omitted from downstream analysis.Last, we used CIH-SNPS to infer ancestry and hybridisation among remaining NB, LRB, and OB.Since CIH-SNPS were optimised to differentiate among basins across the three CIH species(Long et al., 2021), we ran ancestry analysis in Structure for all three species simultaneously.On the other hand, the NewHybridS algorithm uses allele frequency differentiation to infer two 'parent' species from which interspecific alleles are derived in hybrids.Thus, we ran three separate analyses using the full SNP set to infer hybrids of each species pair: NB and LRB, NB and OB, and LRB and OB.Missing genotypes were ignored in all analyses.
v.1.1.2(Jakobsson et al., 2013) using the LargeKGreedy algorithm with 10,000 random permutations.For corresponding analysis inNewHybridS, we classified individuals into one of six hybrid categories (Parent 1 [non-hybrid], Parent 2 [non-hybrid], F 1 hybrid, F 2 hybrid, Backcross to Parent 1 or Backcross to Parent 2).We implemented each analysis in the R package parallelnewhybrid v.1.01(Wringe et al., 2017), running F I G U R E 2 Schematic of hybridization analysis for (a) black bass species (Spotted Bass, Micropterus punctulatus, SPB; Smallmouth Bass, M. dolomieu, SMB; Little River Bass, M. sp.cf.dolomieu Little River, LRB; Ouachita Bass, M. sp.cf.dolomieu Ouachita River, OB; and Neosho Bass, M. velox, NB) in the Smallmouth Bass species complex (SMB-C) and Central Interior Highlands (CIH).Analyses were conducted on three hierarchical levels using three distinct sets of SNP loci: (b) SPB versus all other species in the SMB-C using SPBand SMB-C diagnostic SNPs (SPB-SNPS); (c) SMB versus all other species in the CIH using SMB-and CIH-diagnostic SNPs (SMB-SNPS) and (d) among species in the CIH using SNPs diagnostic for species in the CIH (CIH-SNPS).
using regression analysis in the R package introgress(Gompert &   Buerkle, 2010).For each hybrid individual, we compared hybrid index (proportion of the genotype contributed by one of two parental populations) with interspecific heterozygosity (proportion of heterozygous genotypes in which each allele is derived from a different parental population).F 1 hybrids are expected to have equivalent allelic contributions from both parents (q = 0.5) and relatively high interspecific heterozygosity.Later generation hybrids (i.e.F 2 , F 3 , etc.) and backcrosses vary in parental contribution and interspecific heterozygosity, and evaluating the relationship between these variables allows inference of the magnitude and directionality of gene flow between populations.For each hybrid comparison (SPB vs. SMB-C, SMB vs. CIH, and CIH vs. CIH), all individuals in the dataset were assigned to one of three groups based on identification in NewHybridS analysis:(1) 'Parent 1' (individuals identified as Parent 1 in NewHybridS); (2) 'Parent 2' (individuals identified as Parent 2); or (3) 'Admixed' (individuals identified as either F 1 , F 2 or backcrossed hybrids).We used the prepare.datafunction to compute counts of alleles derived from each 'Parent' population across loci (SPB-SNPS, SMB-SNPS or CIH-SNPS) and then used Bayesian computation with the est.h function to calculate hybrid index.Finally, we calculated interspecific heterozygosity using the calc.intersp.hetfunction.We regressed interspecific heterozygosity on hybrid index and visualised regressions as triangle plots in R.
showed no differential clustering of LD1 loadings between datasets including or excluding missing genotypes for SPB versus SMB-C (Figure S1a,b), SMB versus CIH (Figure S1c,d) or among NB, OB and LRB (Figure S1e,f).

F
Ancestry proportions inferred in Structure for (a) species in the Ouachita Mountains ecoregion using diagnostic SNPs for Spotted Bass (Micropterus punctulatus; SPB) and the Smallmouth Bass species complex (SMB-C; SPB-SNPS) at K = 2 optimal ancestry groups; (b) species in the Ouachita Mountains using diagnostic SNPs for Smallmouth Bass (M.dolomieu; SMB) and species native to the Central Interior Highlands (CIH; SMB-SNPS) at K = 2 optimal ancestry groups; (c) species in the CIH using diagnostic SNPs for species native to the CIH (CIH-SNPS) at K = 3 optimal ancestry groups (deltaK metric) and (d) species in the CIH using CIH-SNPs at K = 8 optimal ancestry groups (Puechmaille metrics).Ancestry proportions are not shown for SPB, SMB or Neosho Bass (M.velox; NB) collection sites for analyses with SPB-SNPS (a) or SMB-SNPs (b), because these samples were pre-screened for non-native SPB or SMB ancestry and were used for assessing structure in the Ouachita Mountains.Thus, ancestry groups are denoted by color to distinguish genomic contributions from SPB and SMB, respectively.Ancestry proportions are shown for collection sites in all three species native to the CIH, including NB, Little River Bass (M.sp.cf.dolomieu Little River; LRB), and Ouachita Bass (M.sp.cf.dolomieu Ouachita River; ORB), for analysis with CIH-SNPS (c and d).Sites are numbered as in Table1; sites within the Ouachita Mountains are labeled in (a) and sites within the Ozark Highlands are labeled in (b).Black arrows indicate sites with relatively high proportions of hybridization and introgression.Colored dotted lines in (d) are drawn approximately around the distribution of each species in the CIH.

4
| DISCUSS ION Our assessment of two endemic, basin-restricted species in the Ouachita Mountains (Little River Bass, LRB; and Ouachita Bass, OB) revealed genetic differentiation and only marginal amounts of hybridisation and introgression with other species in the Smallmouth Bass species complex (SMB-C)-including Smallmouth Bass (SMB) and Neosho Bass (NB)-and with naturally sympatric Spotted Bass (SPB).Low introgression among LRB, OB, and NB suggests that contemporary effects from past stockings are slight.Because stocking TA B L E 3 Hybrids of Smallmouth Bass (Micropterus dolomieu; SMB) and species native to the Central Interior Highlands: Neosho Bass (M.velox, NB), Little River Bass (M.sp.cf.dolomieu Little River, LRB), and Ouachita Bass (M.c.f. dolomieu Ouachita River, OB).
Other ecological and anthropogenic factors, including disturbance associated with human development, may interact with flow regime or act independently to influence interspecific reproductive behaviour or physiology.Changes in the water chemical environment, for example, were shown to significantly impair conspecific mate recognition in female swordtail fish (Xiphophorus birchmanni) via the disruption of chemoreception (odour preference; Fisher

F
Triangle plots depicting regression of interspecific heterozygosity on hybrid index for putative hybrids between (a) Spotted Bass (Micropterus punctulatus, SPB) and the Smallmouth Bass species complex (SMB-C) and (b) Smallmouth Bass (M.dolomieu, SMB) and species in the Central Interior Highlands.Individuals are colored by species native to the CIH (Neosho Bass, M. velox, NB; Little River Bass, M. sp.cf.dolomieu Little River, LRB; and Ouachita Bass, M. sp.cf.dolomieu Ouachita River, OB), and shapes indicate hybrid type (BC indicates 'Backcross').cover in the watershed.Conversely, greater instances of hybrids with the non-native Alabama Bass (M.henshalli) were associated with watersheds with higher proportions of agriculture land cover.Land-use change, such as conversion of forests to agriculture, can increase turbidity in streams (e.g.Tahiru et al., 2020).Elevated hybridisation rates between sympatric congeners may therefore be partially explained by turbidity-induced problems with mate recognition.More research is needed to draw connections between land-use change and turbidity in the Ouachita Mountains, along with experiments to explicitly test the effect of water clarity on reproductive behaviour and hybridisation frequency between native congeners.

Species Population Site Latitude (in/out) Longitude (in/out) n used (n filtered )
Note: n used indicates the number of samples retained for analysis; n filtered indicates the number of samples omitted in filtering steps.Top latitude and longitude coordinates indicate the northernmost collection site; bottom coordinates indicate the southernmost collection site.In sites with only one latitude and longitude coordinate, all samples were collected from a single location.
Note: 'n' is the total number of fish used in analysis.'SMB-C' and 'SPB' indicate number of fish assigning to either parental population.'F 1 ', 'F 2 ', 'BC SMB-C ', and 'BC SPB ' indicate the number of hybrids (H): F 1 , F 2 , SMB-C backcrosses and SPB backcrosses respectively.'Total H' is the total number of hybrids per population across categories, and 'Prop.H' is the percentage of the population inferred as hybrids of any category ( Among individuals identified as hybrids between SPB and SMB-C, hybrid index varied between 0.49 and 0.91, with values closer to 1.00 indicating greater allelic contribution from SMB-C.Interspecific heterozygosity was similarly variable, ranging from 0.20 to 1.00 TA B L E 2 Hybrids of Spotted Bass (Micropterus punctulatus, SPB) and species in the Smallmouth Bass species complex (SMB-C): Smallmouth Bass (M.dolomieu, SMB), Neosho Bass (M.velox, NB), Little River Bass (M.sp.cf.dolomieuLittle River, LRB), and Ouachita Bass (M.c.f.dolomieu Ouachita River, OB).H n Note: 'n' is the total number of fish used in analysis.'CIH' and 'SMB' indicate number of fish assigning to either parental population.'F 1 ', 'F 2 ', 'BC CIH ' and 'BC SMB ' indicate the number of hybrids (H): F 1 , F 2 , CIH backcrosses, and SMB backcrosses, respectively.'Total H' is the total number of hybrids per population across categories, and 'Prop.H' is the percentage of the population inferred as hybrids of any category ( H n